Shifts in biodiversity and physical structure of seagrass beds across 5 decades at Carriacou, Grenadines

Olympus TG4 camera and placed on iNaturalist (www.inaturalist.org/projects/seagrass-bed- flora-fauna-of-carriacou-grenada-2016). In the final compilation of species occurrence for comparison between the 3 sampling years, 69 species or “species groups” could be reliably compared between the 3 sampling years. These include 3 seagrasses, 46 macroalgae, 2 echino- derms, 1 gastropod, 6 corals, 7 sponges and 4 gorgonians. Of these, 56 were identified to the species level. The remaining 13 were classified into “species groups”, which included closely related species that were not reliably or consistently distinguished in the field or for which the taxonomic distinctions are not well defined (e.g., the Porites furcata group refers to P. furcata and/or P. divaricata; Tripneustes ventricosus group refers to T. ventricosus and Lytechinus variegatus). In 1969, 1994 and 2016, samples of seagrass (cut at substrate level) were collected from 3 haphazardly placed 0.0625 m2 quadrats within each 10 x 10 m sampling station. Seagrasses were sorted by species, shaken to remove excess water, and weighed fresh as a measure of above ground biomass. For T. testudinum, leaf width was measured (0.5 mm precision) for 35 haphazardly sampled leaves. Maximum leaf length was calculated as L10, the average length (0.5 cm precision) of the longest 10 leaves for samples of 200 leaves and greater, or as L5%, the average length of the 5% of leaves for samples with less than 200 leaves [49]. In 1994 and 2016, we also counted the number of leaves of T. testudinum in our biomass samples, and the num-

ber of leaves per shoot for 25 shoots sampled in situ at each station. Shoot density was esti- mated by dividing the total number of leaves per unit area by the average number of leaves per shoot. We used a significant linear relationship (F1,30 = 179, P < 0.001) between mass per leaf and the product of maximum length and average width, based on our 1994 and 2016 samples (when all leaves were measured and counted), to estimate leaf number for the 1969 samples from the 1969 measures of maximum leaf length, average leaf width and total fresh weight. We used the average number of leaves per shoot for sites sampled in 1994 to estimate shoot densi- ties in 1969. This assumes that the number of leaves per shoot is similar between years, consis- tent with our observations in 1994 and 2016 (see Results). Ethics statement Aside from sponge samples collected for species identification in 1969, no other physical sam- ples of macrofauna were taken and no permits were required. Throughout the study, sites were accessed mainly by sea and not on privately owned or protected land. No protected spe- cies were sampled, and there was no field or laboratory experimentation. Statistical analyses Differences in frequency of occurrence and biomass of seagrass species across years were com- pared using 1-way ANOVA for each species and response variable. Shoot density and leaf length and width for T. testudinum were compared using 1-way ANOVA for each response variable. Differences in scarp number across wave exposure and year was compared using a

2-way fixed effects ANOVA (car package) with type III sums of squares for unbalanced design. We examined patterns of community diversity using community characteristics (species richness, Shannon diversity) calculated for each station from frequency of occurrence data (proportional frequency over the 12 quadrats) restricted to the 66 species that formed the ses- sile benthic assemblage; mobile species (Tripneustes ventricosus group, Oreaster reticulatus, and Aliger gigas) were excluded. Differences in these response variables across years were tested using 1-way ANOVA. Assumptions of homogeneity of residual variance were evaluated using Levene’s test and inspection of plotted residuals. Where we detected significant main effects, posthoc tests were performed using Tukey’s HSD. PLOS ONE Seagrass bed biodiversity and structure and across 5 decades PLOS ONE | https://doi.org/10.1371/journal.pone.0306897 August 1, 2024 6 / 25